In Western Australia, trans young people can access specialist gender-affirming care through the public health system at the Child and Adolescent Health Service Gender Diversity Service (GDS) at Perth Children’s Hospital. At the GDS, trans young people (<18 years of age) may receive medical gender-affirming treatment with puberty suppression and/or testosterone, oestrogen, and antiandrogens, where appropriate, in keeping with national and international guidelines (21–23). We recruited trans young people age 16–25 who had attended the GDS and had been taking GAH for at least 6 months. Young adult control participants aged 16 – 25 (of any gender identity, and taking no hormone treatment) were recruited by word of mouth, including siblings and friends of the trans participants, as well as researchers’ friends and family members. This study was approved by the Western Australian Child and Adolescent Health Services, Human Research Ethics Committee (#RGS01337).

We acknowledge that individuals not taking GAH may identify as a man, woman, nonbinary or other genders. In addition, not all trans individuals choose to seek medical interventions and take hormones. For our analysis, we have categorised individuals based on whether they were receiving GAH, as well as their sex registered at birth. Therefore, individuals that are not taking GAH we have denoted as “control male” or “control female,” based on their birth-registered sex. The trans groups in our cohort have been denoted as “trans male” or “trans female,” based on their birth-registered sex, gender identity, and the criterion of taking GAH at the time of sample collection (24).

After informed written consent was obtained from the participant (and from parent/guardian of participants under age 18), venous blood was collected in NaHep tubes and processed within 3 hours of collection. For the Max-Par-based experiment, a balanced subset of samples were selected, based on order of recruitment, within a nine-month time period (n = 36). For this subset, 1 mL of fresh blood was utilised for staining as detailed below. For all samples, the remaining blood was spun at 700 g for 10 min and the plasma was transferred to a separate tube, spun again at 700 g for 10 min and stored at -80 °C. The remaining plasma-depleted blood was reconstituted with an equivalent volume of PBS (Thermo Fisher Scientific, USA) and layered onto a Lymphoprep density gradient (Stemcell Technologies) for isolation of PBMC. The isolated PBMC were thoroughly washed with PBS, added to freezing media (RPMI (Thermo Fisher Scientific, USA) supplemented with 50% Hi-FCS and 10% DMSO (Sigma-Aldrich, USA)) and transferred to liquid nitrogen for long-term storage.

Data on GAH regimen, dose and duration was collected at the time of blood collection through a self-report study health questionnaire. Information on recent inflammatory events and administration of anti-inflammatory drugs was also collected. Plasma concentrations of sex hormones including oestradiol (biologically active form of oestrogen) and testosterone were measured by automated chemiluminescent immunoassays using the Architect i200SR platform (Abbott, USA). This testing was performed by PathWest Laboratory Medicine WA, a National Association of Testing Authorities-accredited clinical pathology service ( Table 1 ).

To evaluate individual immune cell profiles, all samples from participants in the cohort (n = 100) were stained with an 18-colour immune cell flow cytometry panel ( Supplementary Table 1 ). For staining, samples were block-randomised, allowing for even distribution of experimental groups in each batch. Cryopreserved PBMC were thawed at 37 °C, and slowly resuspended in RPMI supplemented with 10% Hi-FCS. Cells were washed and resuspended in PBS, and subsequently counted for viability with Trypan blue exclusion method in a haemocytometer. For each sample, approximately 1.5 x 10⁶ PBMC were transferred to a flow cytometry tube, topped up with PBS and centrifuged at 500 g, for 5 min. The cells were resuspended in 100 μL prediluted fixable viability stain (FVS780, BD Biosciences, USA) and incubated for 10 min, protected from light. Following this, 2 mL Fluorobuffer (DPBS (Thermo Fisher Scientific, USA) supplemented with 0.5% BSA (Sigma-Aldrich, USA) and 0.05% NaN₃) was added and incubated for 5 min. Samples were centrifuged and resuspended with 100 μL of the extracellular antibody cocktail ( Supplementary Table 1 ) and incubated for 30 min at 4 °C, away from light. Samples were washed again with Fluorobuffer and fixed/permeabilised using the FoxP3 transcription buffer kit (Thermo Fisher Scientific, USA) according to manufacturer instructions. Briefly, samples were resuspended in 250 μL Fix/Perm and incubated for 30 min at 4 °C. After this, samples were washed in Perm/Wash buffer and resuspended in 100 μL of the intracellular antibody cocktail (diluted in Perm/Wash buffer) ( Supplementary Table 1 ) and incubated for 30 min at 4 °C. Samples were washed with Perm/Wash buffer and resuspended in Fluorobuffer and stored at 4 °C. Acquisition was performed within 2 hours of staining completion on a BD LSR Fortessa X-20, with a maximum event rate of approximately 1000 events/s. Flow data was exported as.fcs files and analysed using FlowJo software version 10.10.0 (BD Bioscience, USA).

For a selected subset of individuals ( Supplementary Table 2 ), fresh blood (as per above) was stained with the Maxpar Direct Immune Profiling Assay (Fluidigm, USA). For this purpose, each sample was incubated with 100 U/mL heparin (Pfizer, USA) for 20 min. Following this, 270 μL of heparin-blocked blood was used to reconstitute each lyophilised antibody panel. The Myeloid and B cell Expansion Panel 2 (Fluidigm, USA) was also added to each sample, and all samples were incubated for 10 min. Following incubation, cells were washed and fixed using the Proteomic Stabiliser 1 solution (Smart Tube Inc, USA) for 10 min. Once fixation was complete, samples were stored at -80 °C until acquisition.

Once all samples in the subset had been stained, the order of samples was block-randomised for CyTOF acquisition. The day before each CyTOF acquisition, samples were thawed at 10 °C and red blood cell (RBC) lysis was performed using the Thaw-lyse buffer (Smart Tube Inc, USA). The cells were then passed through a 35 µm cell strainer and incubated for 10 min. Cells were washed and resuspended in Thaw-lyse buffer and incubated for 10 min. The lysis process was repeated twice, until the cell pellet was white. Samples were then resuspended in 125 nM Cell-ID Intercalator-Ir (Fluidigm, USA) and incubated at 4 °C overnight. On the day of CyTOF acquisition, samples were centrifuged in the Intercalation solution then resuspended in the residual volume. Samples were washed twice in MaxPar Cell Staining Buffer (Fluidigm, USA) and resuspended in Maxpar Cell Acquisition Solution (CAS) (Fluidigm, USA) and counted. Cells were washed again as previously and resuspended at 1 x 10⁶ cells/mL, then stored at 4 °C until acquisition.

Samples were acquired on the CyTOF Helios (Fluidigm, USA) with CyTOF Software. For quality control, CST was performed on the CyTOF Helios using the Tuning solution (Fluidigm, USA) on each day of acquisition. A volume equivalent to 10% of the test volume of EQ Four Element Calibration Beads (Fluidigm, USA) was added to the sample and vortex mixed prior to acquisition. All samples were kept on ice prior to acquisition. Samples were acquired with CAS solution by wide bore injector until the last 100 μL, then resuspended with CAS solution to acquire the entire sample. Files were concatenated for each sample and normalised using the acquired beads prior to being exported as .fcs files.

For flow cytometry and CyTOF experiments, initial data pre-processing for exclusion of dead cells and beads (CyTOF only) was performed in FlowJo ( Supplementary Figures 1 , 2A ). Flow cytometry data was gated manually as per ( Supplementary Figure 2B ). Flow cytometry data and CyTOF data was loaded into in R using FlowCore (version 2.12.0), in R studio. FlowSOM clustering and dimension reduction was performed using the CATALYST package (version 1.26.0) (25). MDS plots were generated to investigate the presence of potential batch effects relating to collection date and acquisition date. Uniform Manifold Approximation and Projection (UMAP) plots were used to guide manual gating of the flow cytometry data.

For analysis of CyTOF data, FlowSOM clusters were generated using markers with stable expression, denoted as ‘type’ markers. FlowSOM meta-clusters were visualised using by UMAP and annotated ( Supplementary Figure 3 ). Individual clusters were exported from R as .fcs files, and the lineage of immune cell subsets was confirmed through conventional gating in FlowJo. For B and T cell subsets, additional sub-clustering was performed following conventional gating in FlowJo ( Supplementary Figures 4 , 5 ). The median fluorescence intensity (MFI) of relevant markers in B and T cell subsets was calculated in CATALYST and exported for further analysis ( Supplementary Figures 6 , 7A, B ).

The plasma concentrations of oestradiol and testosterone were compared between clinical groups using multiple non-parametric comparisons (Wilcoxon method). For cross-sectional analysis, trans groups and matched a no-hormone control group (same birth-registered sex) were compared for cell abundance, and association with hormone concentration by multiple non-parametric comparisons (Wilcoxon method). To determine specific effects of hormones, linear multiple regressions were performed for relevant cell populations, reporting t-ratios. For all statistical tests, p <0.05 was considered significant, and p values <0.1 were reported as trends. All statistical analysis was performed in JMP version 17.0 (JMP Statistical Discovery LLC,USA) and visualised using GraphPad Prism 10.3.1 (GraphPad Software, USA).

To evaluate the immunological impact of GAH, trans participants from the GIM study were compared to control groups who did not take hormones and were registered as the same sex at birth ( Figure 1A ). For all participants, the trans groups were on average 3–4 years younger than the controls, a statistically significant difference. Evaluating the concentration of oestradiol and testosterone ( Figures 1B, C ), levels of oestradiol were significantly higher in trans females compared to control males. There was also a trend towards higher oestradiol levels in control females compared to control males ( Figure 1B ). Testosterone concentrations were significantly higher in trans males and control males compared to control females, and were significantly lower in trans females compared to control males ( Figure 1C ).

Plasma concentration of hormones was measured at one time point for each individual. As per the cross-sectional recruitment of this study, the GAH duration varied from 0.5–4 years. To evaluate if time on GAH impacted plasma hormone levels, hormone levels were compared across time of GAH duration ( Figures 1D–G ). In trans males we observed a significant increase in testosterone concentration between 6 and 12 months of GAH treatment ( Figure 1D ). Other significant differences in hormone concentration were observed in trans groups compared to controls, however there were no additional differences in testosterone or oestradiol levels observed between 6, 12, ≥ 24 month timepoints in trans groups ( Figures 1D, E ).

Regarding endogenous hormone concentrations, in trans males there were no significant differences in oestradiol concentration for any time point ( Figure 1F ). In trans females, the endogenous testosterone concentration was only significantly decreased, compared to controls after 12 and ≥ 24 months ( Figure 1G ).

To determine the impact of GAH on immune cell subsets, the abundance and phenotype of circulating immune cells was evaluated using flow cytometry (n = 100) on PBMC populations and CyTOF (n = 36) on whole blood, ( Figure 2A ). Dimension reduction was utilised to guide manual gating of flow cytometry data ( Figure 2B , Supplementary Figure 2B ) whereas analysis of CyTOF data was based on dimension reduction and clustering, due to the high number of parameters ( Figure 2C ). CyTOF was utilised to identify granulocyte populations not present in the PBMC preparation ( Supplementary Figure 3C ). Abundances of all immune cell subsets detected by flow cytometry or CyTOF are summarised in ( Figure 2D ). Please note, the flow cytometry panel lacked a CD8 antibody, therefore in our gating strategy CD8⁺ T cells were presumed to be present within this gate: CD45^var/CD19^-/CD3⁺/CD16^-/CD56^-/CD14^-/CD123^-/CD11c^-/CD4^-. The presumed CD8⁺ T cells are referred to as “CD4^-” hereafter.

To evaluate the impact of oestrogen-based GAH, immune cell abundance was compared in trans females to control males (both registered males at birth). Based on this comparison, trans females displayed significantly higher proportions of both B cells and CD4^- (presumably CD8⁺) naïve T cells, whereas proportions of eosinophils, CD4^- effector T cells as well as several monocyte and dendritic cell subsets were significantly lower ( Figure 2E ). To determine if the differences were related to administered GAH or changes to endogenous hormones as a secondary effect, the association of cell abundances with plasma concentrations of both oestradiol and testosterone was evaluated using a multiple parameter regression model. From this, only a significant positive correlation of cDC1 abundance with oestradiol concentration was observed, whereas a significant negative correlation of the abundance of B cells and CD4^- naïve T cells with testosterone was observed. A positive correlation of testosterone with monocyte abundance was also observed ( Figure 2F ).

To assess the impact of testosterone-based GAH, immune cell subset abundance was compared in trans males and control females (both registered females at birth). Compared to control females, trans males showed significantly higher proportions of CD4⁺ naïve T cells and CD4⁺ Treg cells as well as a lower proportions of CD11c⁺ B cells ( Figure 2G ). Evaluating the association with plasma hormone levels indeed identified a significant positive correlation between testosterone levels and the abundance of CD4⁺ Tregs, and a significant negative association with the abundance of CD11c⁺ B cells ( Figure 2H ). No association with oestradiol levels was detected in this group ( Figure 2H ).

From the flow cytometry analysis, the administration of GAH was associated with differences in the abundance of B and T cell subsets, as well as DC/monocytes. As the markers included in the immune panel provided limited resolution of B and T cell subsets, these subsets were further evaluated using the CyTOF-based data. Within the B cell population (CD45⁺/CD66b^-/CD14^-/CD56^-/CD3^-/CD19⁺), we identified nine B cell subsets, based on their expression of established B cell markers (26–28), and expression of phenotypic markers. From this analysis, the majority made up of naïve B cells ( Figures 3A, B , Supplementary Figure 4 ).

Comparing abundance of these subsets between trans females versus control males, and trans males versus control females did not reveal any difference between trans females and control males. However, a higher proportion of naïve B cells and CD127⁺ naïve B cells was observed in trans males compared to control females ( Figure 3C ). In the same comparison, a lower proportion of memory B cells was also observed ( Figure 3C ). Evaluation of associations with plasma hormone concentrations identified a significant positive correlation between testosterone levels and naïve B cell abundance in trans males and control females ( Figure 3D ).

To assess associations between GAH and B cell function, we evaluated impact of GAH on expression (MFI) of markers associated with B cell function, such as HLA-DR, CD80, CCR7 and CD40 ( Supplementary Figure 6 ). Although no significant differences in expression were observed for trans males and control females, a trend towards a testosterone-associated increase in HLA-DR expression on both double negative B cells and CD11c⁺ B cells was observed in trans males compared to control females ( Figures 3E, F ). No significant differences in expression of these markers were detected between trans females and control males ( Figure 3G ).

We next evaluated the impact of GAH on T cell subsets derived from the CyTOF-based analysis. For this purpose, CD45⁺/CD66b^-/CD14^-/CD56^-/CD19^-/CD3⁺/TCRgd^-/CD123^- cells were utilised to generate a total of 17 CD4⁺ and 11 CD8⁺ T cell subsets were identified ( Figures 4A, B , Supplementary Figure 5 ). T cell populations were annotated based on their expression of established T cell markers (29–32), and expression of phenotypic markers. The relative abundance of each T cell subset was quantified ( Figure 4C ) and compared for trans males to control females, and for trans females to control males ( Figures 4D, E ). Trans males displayed significantly lower proportions of both CD4⁺ and CD8⁺ memory subsets compared to control females ( Figure 4D ). In trans females, only a lower proportion of Th17 cells compared to control males was observed ( Figure 4E ). Evaluating the association between T cell subset abundance with oestradiol and testosterone concentrations identified a significant correlation between testosterone and CD4⁺ effector memory T cells that were also positive for CD161 in trans males and control females ( Figure 4F ). Although Th17 cells were significantly lower amongst trans females compared to control males, no significant correlation with any hormone level was observed ( Figure 4F ). Finally, the association of GAH with expression of markers associated with T cell function was evaluated. These markers included CD80, CD25, CD57, CD45RA and CD45RO by CD4⁺ T cells and CD8⁺ T cells ( Supplementary Figures 7A, B ). However, no markers were differentially expressed between trans males compared to control females, or trans females compared to control males ( Supplementary Figures 7C, D ).